Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/horizontalmergerOOfish
H331
I
16
.11415
working paper
department
of economics
HORIZONTAL MERGERS AND ANTITRUST POLICY
Franklir M. Fisher
i
No. 439
February 1987
massachusetts
institute of
technology
50 memorial drive
Cambridge, mass. 02139
]S86
HORIZONTAL MERGERS AND ANTITRUST POLICY
Franklin
No. 439
M.
Fisher
February 1987
Draft January 1987
BQBIZ0NT£L_MEBG£BS_MO_MTITByST_PQl,ICY
Franklin M. Fisher
Massachusetts Institute of Technology
antitrust policy
years now,
some
For
mergers has been evolving.
horizontal
towards
It is plain to me that some sort
change was earnestly needed;
of
whether the changes that have taken
place or those that are now proposed by the Reagan Administration
are the appropriate ones is not quite so clear.
the
In
years folowing Bxown_Shge.
ciously
intertwined.
Section
7
These were:
kets or "submarkets"
with
"the
economic
often
some
;
second,
that the definition of mar-
(whatever they are)
is readily accomplished,
The
"market",
product lines and had even
small
a
a
hardly
ultimate
grocery store acqui-
Los Angeles was ruled illegal even though the
had only 7.5% of retail grocery business,
chains and 3,800 stores in operation,
said to be difficult.)
competitive,
of
even if it was obvious that the merger by itself
case in this line was Vc;D_ls_Grocery 3 where
grocery
over-
fraction
(Perhaps the
firms
and
be
firms
could not materially affect competition.
in
mar-
serious
result was that mergers could
successfully challenged if the merging
in their
sition
monopoly
thwart
high-fashion shoe market") substituting for
analysis.
were
lapped
2
to
that
parlance of businessmen ("the Chicago drug-store
the
ket",
the older view
first,
of the Clayton Act was designed
power "in its incipiency"
perni-
two views became
efficiency
merged
there were
and
entry
150
could
Since there certainly can be pro-
reasons for mergers,
2009824
there
are
clear
social costs to having an over-stringent policy.
doctrine.
there is the language of
level,
one
At
to lessen competition or to tend to
substantially
monopoly."
(It
is for this reason that the
More fundamentally,
tion.")
standard,
a
reduce
7
a
pro-
competi-
there are substantive reasons
these are reasons for not
and
create
administration
to change the language to "substantially
poses
such
Section
which speaks in terms of mergers the effect of which "may
itself
be
serious reasons for the "incipiency"
to be sure,
There are,
for
changing
the
language.
In the case of single-firm monopolies, we have had, at least
since the ALCOA case
non-competitive
a
,
legal doctrine that permits us to attack
market structures even if the firm involved
wrongful in 'itself but has deliberately
nothing
done
4
achieve the market structure in question.
There is no
doctrine
Hence,
in the case of tight oligopoly.
has
acted
to
parallel
even where we
are sure that the structure of the market is highly conducive
conscious
likely
parallelism,
to succeed
no antitrust attack on that structure is
even be attempted)
(or
explicitly
Only
.
implicitly collusive acts can be successfully attacked,
a
win
that
by the Government will leave in place the very
makes
.
occur again.
to
it likely that
similar anticompetitive
or
and even
structure
events
will
7
Of course,
it will sometimes be the case that no structural
remedy for tight oligopoly is possible.
Just as there are natu-
ral monopolies,
there are natural oligopolies,
inevitability of
a
monopoly-like structure is
defense under Section
2
of the Sherman Act,
(or
and, just as the
ought to be)
a
so the inevitability
of
oligopoly structure ought to be
tight
a
structurally-oriented anti-oligopoly act.
a
defense
under
a
But at present that is
not an issue, because such an act does not exist.
The closest thing that we do have to such an act is
Where
7.
merger or a series of mergers will result in
a
Section
oligopoly structure.
us
"inevitability"
structure can be most easily examined by weighing the
that
of
pro-
anticompetitive effects of the merger in the context of
and
pre-merger situation.
is
the move to a tight oligopoly
be halted at a time when the
can
tight
a
with its present language permits
What is more,
to prevent it.
structure
7
Section
Finally, since, as
generally far easier to enjoin
divestiture afterward,
a
the
practical matter, it
merger than to order complex
a
there is something to be said for dealing
with structural oligopoly "in its incipiency" by dealing with the
mergers that will create it.
There
which
are,
however,
problems here.
Consider the case in
tight oligopoly structure will be
a
series of mergers if they are not stopped.
dealing
with
because
in
other
hand,
itself innocent and
even
it may stop a merger
pro-competitive
of a line of mergers that may never happen.
if
di-
On the one hand, if antitrust attack
simply
the Court or the Justice Department envisages it as
forerunner
a
In such a situation,
begins with the first merger in the series,
is
through
the problem through merger policy rather than
rectly runs certain risks.
which
attained
antitrust attack waits for later
anticompetitive effect seems certain,
mergers
one can regard the
cipants in those merger as being treated unfairly.
On
the
the
when
parti-
Further, such
a
policy can provide an incentive to merge while the merging
structural anti-oligopoly policy,
A
good.
by contrast,
—
involve all the participants in the oligopoly
those involved in
As
direction
the first problem
of
—
rather than only
as defendants.
has
erred
like them would together be anticompetitive.
gers
the
in
mergers
attacking particular
-
would
the Kantian categorical imperative shows that many
because
the incipiency doctrine too far.
of
—
merger policy
turned out,
has
it
single transaction
a
mer-
This carries
On the other hand, the attempt
administration to change the language of
the
is
Section
7
to
"substantial reduction in competition" will lead to the
require
a
second
problem
•
—
otherwise
attacking only later mergers in an
symmetric series.
There is no solution for this in the realm of merger policy.
The
problem
arises
substitute for
because we attempt to
make
structural policy towards oligopoly.
a
policy
merger
It is
at
o
best an imperfect substitute.
pursuit
The
of
presupposes
however,
structures
when
any structural policy
that
we
we see them.
direct structural policy,
can
oligopoly,
towards
recognize
anti-competitive
That would be a requisite
and it is crucial for the
more indirect approach embodied in merger policy.
that requisite is not easy to meet,
a
necessarily
Unfortunately,
and there is a temptation to
avoid difficult analysis in favor of easily measured,
rect standards.
of
but incor-
While some progress has recently been made, this
problem pervades all aspects of merger policy.
The
market
area
in which some progress has been made is
definition,
that
of
in which the Department of Justices' s Guider
lines
9
have focussed on the right sort of things.
definition is an artificial problem created by anti-
Market
the
For any other purpose of economic
litigation.
trust
analysis,
binary question of whether particular firms or products
or "out" of a given
"in"
market is
a
antitrust cases, that question is not
results
a
whose
those
The proposition that flexible wrap-
papers substitute for cellophane at
ping
in
constraints
the
firms and products put on the power of
actions are being examined.
Even
useful one if substantive
What matters are
turn on the answer.
other
that
meaningless one.
are
a
high cellophane price
but not at lower ones already contains a good deal of information
concerning the ability of
monopoly
a
sole supplier of cellophane to charge
There is nothing to gain and much to lose
by
the Procrustean device of summarizing that information either
in
the
prices.
that flexible wrapping papers and cellophane
statement
"in" the same market
or
in
are
(the Supreme Court's position in Cellppbaog)
the statement that they
are not
(the
position
many
of
commentators)
Such Procrustean activity, however, has historically been of
overwhelming
definition
importance in antitrust cases.
being used as
information,
merger
market
device for summarizing and organizing
has become the principal issue.
To return
Nestle's
of Stouffers in the mid-1970s to know that
analysis
that case was not helped by a debate over whether there is
1
"market" consisting of high-priced, frozen, non-ethnic entrees.
If
to
one need know little about the facts of
cases,
acquisition
in
it
a
Instead of
market
definition is to be at all useful
in
a
11
antitrust
cases, it must be the beginning, rather than the end of analysis.
Market
definition ought properly to define the universe of
dis-
course
within which analysis will take place.
that
This means
the "market" should include those firms and products necessary to
of the pricing and product
understanding
the
behavior
the
in
An alternative way of putting this is to say that the
"market".
must include those firms and services that act to
"market"
con-
strain the activities of the firm or firms that are the object of
That such constraints may not all be equally powerful
attention.
points
merely
to the facts that analysis does not end when
market has been defined and that simple-minded measures of
or concentration,
definition,
the
power
like simple-minded binary treatments of market
unlikely
to be adequate.
return
to
this
Department of Justice (DOJ) Mg£ggr_GuidgliD£s
(in
both
are
I
12
U
T
below.
The
the 1982 and 1984) versions, are
firms that could,
aside
above
if they colluded,
profitably raise prices five
Putting the
current levels for a year.
a moment,
for
major step in the direction of
DOJ defines a "market" as the minimum collection of
sanity here.
percent
a
this is plainly the
right
details
approach.
If
prices cannot be so raised, then supply or demand substitutability
puts
important
included.
be
constraints on the power
of
those
To leave those constraints out of the "market"
to have much of the action take place off stage.
already
would
The Guider
lines implicit focus on constraints as the principal question
to
be asked in market definition is absolutely right.
It
is
less clear that the specific details of
liDgsJ. approach are correct.
the
Guide;:
Is five percent the correct
thres-
amount
hold
amount
to use for price rises?
of time?
Is one year
the
correct
Should the five percent be applied to
current
price levels or to something else?
begin with the third of these questions,
I
since the answer
to it has implications for the answers to the first two.
The use
of current levels as the base for the test is consistent with the
view that merger policy is a preventive, designed to keep matters
from
getting worse.
that
merger
policy is
a
substitute for
view
if one takes the
On the other hand,
structural policy
a
to-
wards oligopoly, then one may want to use competitive rather than
current
levels as the base.
duopoly
in which prices have already been raised to
Consider an already
percent of the level at which
product
(flexible
Guidelines
a
tight
fairly
within
five
competitively-produced substitute
wrapping papers) can
compete.
present
The
since the market
would let the duopoly merge,
have to include the producers of the substitute.
would
Since conscious
parallelism may not always be easy to maintain, permitting such
merger
last.
may
(A
make permanent
a
situation that otherwise might
a
not
similar statement applies where there are more than two
firms in the original oligopoly.)
Further,
since the merger
of
the two duopolists is not "economically inevitable" or solely the
result of "superior skill,
the
foresight,
and industry,"
13
there is
possible anomaly that the Guidelines used in the administra-
tion of the supposedly more stringent Clayton Act, would permit
merger
leading
to a monopoly that could
then
be
a
successfully
challenged under Section Two of the Sherman Act.
On
the
other hand,
while the use of current price
levels
implies an acceptance cf existing power and behavior,
competitive levels has its own problems.
what competitive prices would be,
just
prices
used as the base,
were
the use of
It is not easy to know
in
practice.
there would be
danger
serious
a
such
If
that DOJ would simply look at profits or profits-sales ratios
a
mechanical attempt to compute competitive prices.
be
This would
both because profits play an important
mistake,
a
role
competitive industries and are not absent save in long-run
librium,
accounting
because
and
measures of
profits
knowing
14
what competitive price levels are is more apparent
decide
whether
than
market participants can raise
prices
above
competitive levels is precisely the kind of analysis required
do
sensible job of market definition by considering the
a
straints on behavior.
to
con-
It is not particularly different in
kind
the qualitative analysis now required by the use of current
from
levels as the base.
to
in
The kind of qualitative analysis required
real in this context.
to
the
or
the difficulty
however,
in
equi-
profits-sales ratio do not tell one what one wants to know.
If that danger can be avoided,
in
Only if detailed quantitative analysis were
performed would the exact location of
be
levels
matter,
and
such
competitive
analysis is typically
not
price
practical
anyway.
As
the question of whether five percent
this suggests,
the correct figure for the test may not be very
five
The
important.
percent figure does serve to focus attention on the sort of
effects
serves
is
that will be considered important,
only
to
give
a
but beyond
spurious impression of precision
analysis that is generally imprecise.
that
it
to
an
On the other hand, quanti-
tative analysis need not always be impossible,
point
so there is
some
in considering whether five percent is a reasonable figure
to use.
The answer here depends on the costs and benefits
By
using
a
high figure,
one allows mergers to slip by that
lead to elevated prices and welfare loss.
using
low figure, one runs the risk of prohibiting mergers that
a
ther,
may
On the other hand, by
relatively harmless and may have efficiency
are
involved.
reasons.
Fur-
bears the cost of administrative or judicial proceed-
one
ings in order to stop a fairly small harm.
Since it is impossi-
ble to decide where to draw the line without
a
detailed
analysis
of the likely welfare losses and gains in each case, five percent
seems to me to be
however,
price
I
sensible administrative rule.
a
As indicated,
would apply it to competitive rather than to current
levels,
which suggests
a
more stringent rule for
already
non-competitive industries than DOJ now uses.
take a different view on the use of one year as the
I
criterion for the test.
which
entry
short
a
not
a
me
Together with two years as the time
will be considered,
this seems to me to imply
time horizon and too restrictive
the
power
involved
in
Self-correction within that time seems
to
test.
particularly
about,
to raise prices by five percent
everybody
if
colludes does not imply that prices will in fact be so raised.
would
market
be inclined to use
definition
and
too
is
to make the problem not worth bothering
since
in
Two years
a
particularly long time compared to the time
litigating a merger.
time
a
longer time,
perhaps two
four years for ease
of
years
entry.
I
for
Again,
it is hard to be sure.
however,
Since market definition is the
threshold event for analysis, a more stringent rule may be appro-
priate
in deciding
whether to investigate further than in
deci-
Certainly, a more stringent
ding whether to challenge a merger.
rule is appropriate in that circumstance than in setting a
stan-
dard for judicial decisions.
This brings me to an important point.
between
deciding
cases to investigate or oppose,
what
—
difference
a
guidelines
as
to
and deciding on a judicial
Arbitrary rules are inevitable and may even be useful
standard.
in
on guidelines for triage
There is
the first context.
Department
They are
a
menace in
the
The
second.
of Justice has not always recognized the
difference,
particularly when it comes to the use of concentration measures.
principal reason for such unwise concentration on
The
what
to be the non-question of market definition is that things
ought
"in" the market will be counted in measures such as market shares
or concentration indices which,
answers
give simple
it is vainly hoped,
to questions of market power or the likelihood of
competitive activity.
Things "out" of the market,
anti-
on the other
hand, play a much more minor role, often coming in, if at all, in
terms of
The
makeweight reference to ease of or barriers to
a
problem
is
that the use of such indices does not
entry.
fact
in
produce correct results.
The
fact is that the analysis of oligopoly does
not
yield
useful results relating structure to conduct and performance.
know
in a general way what the features are that make
parallelism
firms,
more or less likely
conscious
(number and size distribution of
complexity of the product,
10
We
etc.).
Unfortunately,
such
is nowhere near precise enough to substitute
knowledge
study of specific situations.
while it is clear
In particular,
conscious parallelism is more likely the smaller the number
that
of
the
for
firms,
whether
is not true that we have any serious idea
it
danger point is reached at four firms
the
rather
than
five or, indeed, what the function in question lokks like.
while
likely
the more concentrated is the market,
reason
for
Hirschman
it seems clear
picking
Index
(HHI)
Simi-
that conscious parallelism is
larly,
out particular levels
as danger points.
there is
of
more
sound
no
Herfindahl-
the
Indeed,
to
as
while the HHI
itself seems a reasonable way to measure concentration,
there is
neither theory nor reliable econometric evidence showing that the
HHI is a sufficient statistic for the effects of concentration on
non-competitive behavior.
It
would
15
therefore be
a
great mistake if courts
were to adopt the practice of judging mergers by
gress)
or even primarily at pre- and post-merger HHI
only
with market shares in monopoly cases,
crudest
Con-
(or
looking
levels.
As
the HHI provides only
the
of indications as to what we want to know.
Any serious
merger case must ask specifically about the possibility of
collusion.
This means an investigation of the particular situa-
involved
tion
tacit
including
not necessarily
(but
limited
to)
a
serious analysis of ease of entry.
Such strictures, however, do not apply to the use of the HHI
for
administrative purposes as in the DOJ Guidelines.
mited resources and finite time,
decide
somehow
the antitrust authorities
what cases to investigate and then
11
With
pursue.
li-
must
If
decision is net itself to require
that
In that circumstance,
readily.
investiga-
rules must be used that can be applied
some
then
tion,
full-dress
a
fairly
the use of the HHI to trigger or
turn off investigation appears warranted.
Since
would only use the HHI in this way,
such
to treat foreign competition recede in
how
as
I
exist
quotas
or are likely,
questions
importance.
then the availability
If
foreign
of
capacity does not put the same constraint on post-merger anticom-
petitive
behavior than would the same capaicity
domestic hands.
independent
in
would calculate the HHI both including foreign
I
production beyond the quota level and excluding it.
a
difference,
"right"
mistake
in
then there is something to analyze and investigate
There is no point in wasting time arguing which one is
further.
the
If it makes
computation'.
As with market definition,
it
deci-
a
that foreign competition is either the same as domestic
sion
not present at all.
point
a
to suppress the fact that foreign competition may matter
different way from domestic competition by forcing
a
is
of
a
or
Calculation of the HHI ought not to be the
merger analysis,
but only
a
signal
for
further
investigation.
Are
the
levels currently used in the Guidelines, the
How can one know?
Plainly, a very
post-merger HHI makes it most unlikely that
anticompetitive
ones to use as such signals?
low
behavior will
merger
(or
can)
result from the merger.
Plainly also,
that raises the HHI by a very large amount and leaves
very high is one that requires investigation.
low"
right
and "very high" mean?
it
But what do "very
Is the 1800 cut-off the right
To know this with much certainty would be to know what we
12
a
one?
empha-
do not know
tically
—
exactly how the HHI relates to non-compe-
titive behavior
One can get a little farther than this, however.
of
The danger
setting the trigger levels of the HHI too high is that
mergers will slip through.
competitive
them
setting
with
beset
effects
been
dangers
of
too low is that the antitrust authorities will
be
One of the
many cases of high HHIs with
claims
experience
in this regard under
the
What has
.
Guidelines?
that the call for DOJ to consider other
suspect
offsetting
of
(and may lose such cases if they go to trial)
the
anti-
things
I
besides
concentration suggests that the trigger levels are set low rather
than
particularly because my experience suggests that DOJ
high,
has a strong tendency towards too narrow a market definition.
This
is
not necessarily a bad thing.
setting such levels is to trigger investigation,
better to waste resources on an investigation of
shown
to
purpose
of
it may well
be
If the
a
be harmless than to fail to investigate
merger that is
a
that
merger
will turn out to be harmful.
Unfortunately, there are other costs to
Mergers
are
sometimes
delicate
creatures,
litigation can be extremely expensive.
Guidelines
levels.
mergers.
used
by
low trigger levels.
and
antitrust
The HHI levels set in the
such
can therefore act to deter mergers that involve
Setting
the
levels
low
can
deter
socially
useful
This is particularly likely if the trigger levels
the
authorities not as signals to investigate
signals to oppose.
is a natural,
Alas,
this is likely to be the case.
if distressing
are
but
as
There
tendency for the Antitrust Division
13
to become fascinated with its own Guidelines,
and to
focus
on the HHI levels mentioned therein as though failure to pass the
tests of the Guidelines were proof that
tive rather than merely being
analysis
further
Unfortunately,
merger was anticompeti-
signal for further analysis.
(but not necessarily)
primary
The
a
a
the only
the investigation of
is
item
barriers
such
in
entry.
to
this is generally recognized in principle,
while
there is mass confusion over what it involves in practice.
analytic use of the term "barriers to entry"
The
can
from working."
only be
entry
a
must
as
"Barriers to entry prevent the competitive
part of the sentence:
process
comes
Similarly:
competitive result."
there
"Where entry is easy,
Accordingly,
be something that interferes with
must be something that allows incumbent firms,
barrier
a
to
competition.
if they
It
collude,
to charge non-competitive prices and earn supra-normal profits.
follows
It
not everything that
that
makes
difficult or uninviting is necessarily a barrier to
mere
built
mies
or
necessity of building
a
of scale can be)
creating
a
The
entry.
already
plant when incumbents have
theirs is not such a barrier
appear
entry
(although associated
econo-
Neither is the necessity of advertising
.
reputation automatically a barrier.
long-run
the
phenomenon
involved must give incumbents
advantage,
permitting
them to earn supra-normal profits on
rier,
whole
bar-
To be a
a
the
process of getting into the market and continuing to
without inducing others to enter and bid those profits away.
This
Antitrust
is not an easy
Division
concept to apply in practice,
does not have a good track record
recent example will serve to illustrate the point.
14
1
act,
17
and
here.
the
A
In the recent
was
Northwest-Republic airline merger
witness for Northwest)
a
market.
,
which
(in
DOJ took quite a narrow view of the
addition to the position discussed in footnote
In
I
16,
above, DOJ argued that air passenger traffic on routes out of the
merged airline's Minneapolis hub,
peted
could only be effectively com-
Minneapolis.
by another airline also having a hub at
for
This is not
a
compelling position.
What keeps an airline with
a
hub at Denver from competing on equal terms with one at Minneapofor traffic between the two cities?
lis
with
hub at Dallas,
a
say,
Why cannot an
airline
and already serving cities
between
and Minneapolis simply extend its flights to compete
Dallas
between Minneapolis and those other cities?
traffic
these
questions
entry
and
assume
aside in order to concentrate
arguendo that
a
on
But
I
barriers
hub at Minneapolis
would
for
put
to
be
necessary to compete with the merged airline.
There
were
structing such
gate
a hub.
facilities.
already
no obvious barriers to another
Landing slots were not
Further,
of
problem, nor were
since both Northwest
had hubs at Minneapolis,
economies
a
con-
airline's
Republic
and
it could not be the case
scale made the possession of such a hub
that
natural
a
monopoly.
Why then did the DOJ claim that there were barriers to entry
and
go on to oppose the merger?
Division,
It
is not an
said
the
attractive to build
a
and other airlines
a hub.
between
will
not
hub there in the presence of
the
East and West Coast cities,
it
Antitrust
attractive place to have
is too far North to be an efficient connecting point
major
find
Minneapolis
Because,
15
large
number of flights "controlled" by the
west.
Indeed, DOJ conducted an informal survey of other airlines
post-merger
told them that they would not regard hubbing at
who
North-
Minneapolis
as an attractive post-merger proposition.
This position misunderstands the proper analysis of barriers
issue should have been whether
The
entry.
to
other
find hubbing at Minneapolis attractive if the
would
ppstrmerger
Whether
Northwest sought to raise prises and reduce output.
not
airlines would find Minneapolis attractive
other
aggressively
Northwest
post-merger
airlines
competing by
with
the
offering
the
service previously flown by the two merger partners and doing
pre-merger price was irrelevant.
the
at
Even
or
so
obviously
more
was the issue of whether Minneapolis is inherently an
irrelevant
The geographical position of Minneapolis is not
attractive hub.
something that gives incumbents an advantage over entrants.
Having
said
this,
I
cannot forbear adding
that
DOJ
factually wrong about the attractiveness of Minneapolis as
In fact,
jection
great
because the earth is
sphere,
a hub.
the usual Mercator pro-
of North America gives a quite misleading picture.
circle
and that,
The
pass
quite
together with prevailing
winds
routes from East to West Coast cities
close to Minneapolis,
traffic patterns makes it the second most attractive hub for
and
flights,
such
DOJ
a
was
'
s
hearing
position
a few
minutes worse
than
Chicago.
Shockingly,
on this indisputable matter persisted after
in the case.
At discussions following
those
the
hearings,
higher-ups in the Division made the same argument about Minneapolis's
wrong.
position and were quite surprised to learn that they
The symbolism is clear.
I
16
were
fear that, at least as regards
the
the Department
of barriers to entry,
analysis
Justice
of
still believes that the earth is flat.
This is
defined
Particularly if markets are to be narrowly
pity.
a
would put great weight on it in considering
Having said this,
I
must go en to caution against
permit
the HHI to serve as more than
produce
a
useful quantitative index of ease of entry.
the antitrust authorities
quantitative
need
so also we know too little to be able
premature attempts in this regard.
on
not
rough signal of the
may change as the science progresses,
this
for
a
of
measure
Just as the state of our knowledge does
for further investigation,
to
attempts
important question by the creation of some summary
ease of entry.
I
prospective merger.
a
avoid what ought to be a thoughtful and detailed analysis
this
of
set
the analysis of entry is absolutely crucial.
relatively low,
to
investigation
HHI levels that trigger further
and
standards
is well to
it
There is
While
temptation
great
a
(and perhaps the courts)
as a substitute for
avoid
real
focus
to
analysis.
Economists ought not to offer such temptation unless the delivery
soundly backs up the promise.
Another
example drawn from the Northwest-Republic
illustrative
here.
likely to enter
a
Correctly
observing that airlines are
service,
for example,
other origins at Kansas City)
lihood
of
constructed
entry
is
more
given city-pair route if they have traffic that
feed into that route (more likely to enter Kansas
lis
case
City-Minneapo-
if they can collect passengers
,
DOJ introduced
called the "feed ratio."
as follows for a given city pair,
17
a
from
measure of like-
That
measure
A and B.
was
Assume
first that there is only one incumbent airline and one
potential
The "feed ratio" is the ratio of the sum of the poten-
entrant.
entrant's emplanements at A and B divided by the sum of the
tial
incumbent's emplanements at the two cities.
than one incumbent-entrant pair,
the "feed ratio" is taken to be
the
maximum over all such pairs of this ratio
DOJ
argued
least at first)
(at
Where there is more
that the fact
of
emplanements.
that
the
"feed
was relatively low for a number of routes involved in the
ratio"
merger showed that entry was difficult.
This is, of course, nonsense.
lis city pair,
those
airline.
be flying on
by definition,
must,
between
incumbent
an
Other passengers emplanng at those cities are certainly
Of what possible' relevance to a decision
somewhere else.
going
Passengers wishing to fly
for example.
cities
Consider the Boston-Minneapo-
by Delta to enter Boston-Minneapolis service is the fact that
my
wife, who has no reason to travel to Minneapolis, sometimes flies
from Boston to Cincinnati to visit her parents?
It
predict
is
entry and not surprising that the chief
actual
DOJ eventually admitted that it was not an entry
for
What is disturbing is that DOJ made
it
forward.
I
.
putting
That lure should always be resisted.
After the decision to investigate
on the basis of the HHI,
a
proposed merger has been
ease of entry is the
that should be investigated first.
the
witness
predictor.
fair production of
a
phenomenon
That is so, first because of
intrinsic importance of the role of potential entry (or
lack thereof)
to
believe this was because of the powerful lure of
spurious measurability
taken
fails
thus not surprising that the "feed ratio"
and,
second,
because
18
a
the
finding that entry is easy
Where entry is not easy or the issue in
should
be dispositive.
doubt,
further investigation must be undertaken as to the likely
effects
the proposed merger on
of
competition
and,
those
if
effects are found to be negative, on any offsetting efficiencies.
gone beyond an analysis of entry to an
Having
merely
first as a threshold matter.
would
I
those other factors as tie breakers.
use
concentration
One
of
not
One
analyzes
then
analyzes
because it may dispose of the question if the answer comes
entry
out
bearing on effects on competition,
factors
other
analysis
particular way.
a
have
But one must not forget that we
do
not
good enough theory of oligopoly to be able to infer anti-
a
competitive results from structure in any precise way.
Instead,
analysis should always bear in mind that the question
merger
at
issue is the likelihood, or at least the ease, of anticompetitive
The complexity of the product, the extent to which an
behavior.
effective
many
agreement would require implicit
tacit
negotiated
price,
transaction
prices instead of
collusion
single
a
list
the ease with which cheating on a tacit agreement can be
detected,
analysis
and similar matters are properly
these
once
subjects
appears that concentration will be
it
entry difficult.
high
for
and
It is not unnatural that the burden of proof in
matters will devolve onto the participants in the
such
on
prospec-
tive merger, but that is not to say that such matters should only
come into play in otherwise doubtful cases.
The
burden of proof as to cost savings or other
efficiencies
here
I
whould also be on the proponents of
would require
a
very high standard.
19
a
offsetting
merger,
but
That is because such
claims are easily made and,
I
often too easily believed.
think,
Two examples will illustrate this.
General Motors and Toyota proposed
When
assemble
small car in California,
a
joint venture
a
one would have thought
considerations would have prevented it.
antitrust
that
Here the two
largest
automobile manufacturers in the world were combining
produce
a
provide
an
obvious
prices.
Even
question
by
prices,
The price of that vehicle
vehicle.
reference point for the
very
to a particular
sion approved the joint venture.
argument that the venture
the
likely
setting
average
other
of
would
realize
efficiencies,
Japanese
Presumably,
GM would then be able to use those secrets in other plants.
is far from clear that the
by
secrets
all.
at
hardly
It
to
be learned,
labor
relations
and
took a joint venture to
learn
about
those.
the extent that there were production "secrets" to
it seemed unlikely that GM would learn very much from
assembly
Finally,
to be
at least,
management were believed to be the source of efficien-
Moreover,
an
The Japanese system of
Japanese
so-called
were not believed by Ford officials,
inventory
cies.
The
19
accepted
efficiency argument
the FTC was more than superficial.
"secrets"
to
because
It did so principally
automobile manufacture more efficient than American.
It
car
Yet the Federal Trade Commis-
GM would learn from Toyota the secrets that made
since
to
other
of
use of a particular average seemed likely
facilitate conscious parallelism.
of
was
to
though GM and Toyota proposed to set the price in
reference
the
to
plant
when the engines
were
produced
in
Japan.
GM already had relations with other Japanese automobile
manufacturers
20
All
by
in all,
the FTC appears to have been too easily swayed
the difficulties of the American automobile industry and
success of the Japanese.
It accepted fairly superficial promises
of
efficiencies to be gained and approved an arrangement
to
be
anticompetitive.
result
of
Ford,
a
One need only contemplate
similar application for
example,
for
to
the
likely
likely
the
joint venture by
GM
and
realize the tremendous weight that
the
a
efficiency argument was given.
My second example relates once again to airlines.
Here, the
Department of Transportation (DOT) has approved a whole series of
mergers.
On
the whole,
I
regard those approvals as warranted.
The entire process of airline deregulation rests on the view that
city-pair
and
"markets" are contestable.
other facilities are available
number
large
of airlines)
,
So long as landing
(or can
there is
a
be purchased
slots
from
strong presumption
mergers of domestic airlines cannot result in much market
That
presumption,
however,
where entry is in fact difficult,
does not extend to
DOT's
that
power.
situations
automatically
and it does not
extend to acquisitions involving foreign routes.
a
In particular,
approval of United Airlines acquisition of Pan
American's
Pacific Division is open to very serious question. 20
.
Entry
Asia
is
service
plays
ding
.
into air transportation between the United States and
far from easy.
This is particularly true
between Japan and the United States,
a
and
as
regards
that
service
vital role both because of Japan's importance as
partner and because of its geographic position.
tion does not apply to that service,
21
and,
indeed,
a
tra-
Deregulathe Japanese
historically reluctant to
been
have
there
Further,
are
considerable
permit
expanded
restrictions on
service.
the
use
of
Tokyo's Narita airport.
Before
the acquisition,
quite
concentrated
would
permit
percent
The
2542 in 1984)
the number four carrier
service
(United)
with
,
about
Lines and Northwest) each had
Air
a
bit more
7
carrier
Numbers one and
which had about 19 percent.
was
acquisition
The
.
the market to combine with the number three
of
(Pan American)
(Japan
(an HHI of
mainland
Japan-U.S.
than
two
30%.
acquisition (in terms of 1984 figures) caused an increase in
the HHI from 2542 to 2812,
DOJ's Guidelines.
well beyond the trigger levels set in
21
Before the acquisition, there was substantial price competition
in particular,
United,
acquisition,
traffic
it
Apollo computer reservation system,
be
the
only
airline providing both
After
attract
could
first by manipulating
competing on price,
without
United
seemed likely that
actively
had
gateway.
to increase traffic through its Seattle
sought
the
various kinds.
.of
its
and second, because it would
a
really
extensive
route
structure in the United States (acquired during regulation) and
large system of routes connecting at the Tokyo hub
—
a
latter
the
being Pan American's legacy from regulation.
There
service
is
meant
no doubt that the provision
a
real benefit to pasengers.
regarded this as an efficiency.
of
such
DOT very
integrated
properly
What is not so clear is whether
that efficiency should have justified the acquisition.
Had
have
the acquisition not been approved,
either
sold its Pacific Division to a
22
Pan American
different
would
domestic
airline not already serving Japan or else would have continued to
(The Pacific Division was
operate it itself.
before
the
profitable,
Pan American had announced
acquisition,
Pacific service for the summer of 1985.)
expanded
and,
plans
for
In the latter
Pan American would certainly have continued its program to
case,
expand its domestic route structure.
quisition
would
have
not
forbidding the ac-
Hence,
prevented
service
integrated
from
developing
American was not the only airline that would have deve-
Pan
loped such integrated service.
developed
its
own Tokyo hub,
domestic route system.
largely
sion,
gradually
which had
was also striving to
expand
its
(After the transfer of the Pacific Diviof the need to catch up with
because
acquisition United,
Northwest,
'
post-
the
Northwest strove to expand quickly by acqui-
ring Republic.)
United itself could have expanded.
Most important,
was already creating
a
rival hub at Seoul.
into Japan was difficult,
States
Further, while entry
it was not impossible,
and the United
government could have made expansion by United
object of negotiations with Japan.
This was
United
a
a
primary
very real possibi-
lity, because the Spring of 1985 saw an agreement between the two
countries to open as many as three new routes.
United could have
been given those routes.
In short, absent the acquisition, there might well have been
three
companies competing to provide
acquisition
service.
that number to no more than two.
reduced
connection, DOT took
integrated
a
In
The
this
very limited view of its responsibilities,
23
for
refusing,
to connect the award of the new
example,
with the outcome of the case.
with
would
It appears to have been impressed
post-acquisition
irrelevant argument that the
the
be
stronger competitor than
a
routes
United
pre-acquisition
the
Pan
American.
because
of
efficiency
considerations if the efficiencies involved could
be
obtained in
a
this suggests,
As
I
would not approve mergers
Further,
less restrictive way.
I
would hesitate to
efficiencies as an excuse for permitting
merger
if
those efficiencies are unlikely to be passed on to customers.
In
use
the
such
Pacific Division case,
for example,
a
the benefits of
inte-
grated
service could have been achieved while maintaining compe-
tition.
That would have ensured that the travelling public would
benefitted from those efficiencies without paying more
have
for
The approval of the acqui-
them in the form of increased prices.
sition created the efficiencies but also made it very likely that
all benefits would be captured by United itself.
am,
I
payments
concerned
obtained
who
sensible of the argument
ought not to matter to economists,
be
only
of course,
outweigh
with the question
but hard to evaluate.
achievable
mergers,
nally,
whether
In practice,
however,
I
of
not
would
otherwise
Efficiency arguments are easy
to
The same efficiencies will often be
in less restrictive ways,
on the other hand,
a
should
efficiencies
mergers for efficiency reasons if they seemed
likely to be anticompetitive.
make,
so that one
deadweight loss and not with the question
captures the savings.
approve
of
transfer
that
have
a
particularly if one waits;
way of being permanent.
policy of approving anticompetitive mergers for
24
Fieffi-
reasons is likely to promote a dissipation
ciency
resources
of
into rent seeking.
All of this, however, supposes that
been
has
a
proper merger analysis
carried out and the proposed merger found to be
competitive.
I
would certainly accept evidence of
showing that the merged enterprise will be
as
tor.
efficiencies
tougher competi-
merger analysis continues to be dominated by the
If
surement
a
anti-
of concentration,
I
weight
would put considerable
a showing as offsetting the really crude presumption
such
meaon
resul-
ting from market definition and the HHI.
To sum up then,
I
think the DOJ Guidelines are roughly right
properly interpreted.
if
however,
Such proper interpretation,
requires the use of market definition and the HHI only as signalling the necessity of serious analysis.
done,
then
pre-merger screening can serve an important useful purpose,
pre-
venting
If that is
lengthy litigation to force the disgorgement of
already
digested assets.
On the other hand,
there has been
a
tendency for the Guide;:
lines to substitute for analysis, with DOJ focussing on issues of
market definition and concentration measures.
the present
In
rious
in
a
(and likely future)
That is
state of our
a
mistake.
knowledge,
analysis of market power and oligopoly cannot be
se-
subsumed
few spuriously precise measurements.
The Reagan administration has generally been very permissive
in its merger policies.
viewed as
HHI
a
To an extent, that permissiveness may be
correction to the tendency of DOJ staff to substitute
measurement
for economics,
25
but that is only
true
if
one
thinks
fact,
opposed
of different mergers as substitutes for each
mergers
by
standards,
have sometimes been wrongly blocked (or at
DOJ) because of unthinking application of
and
sometimes wrongly approved because of
find efficiency excuses
(a
The
a
least
Guideline
a
wish
to
United-Pan
two mistakes do not compensate for each
and neither approach is
In
wish that may be greatest where compe-
tition with foreigners is involved as in GM-Toyota or
American).
other.
substitute for sound analysis.
26
other,
FpptDOtes
BrpWD_Shpe_CpmpaDy v
1.
"incipiency" doctrine goes back
The
2.
discussion
of the original Clayton Act.
63rd Cong., 2nd Sess.
698,
used when Section
81st Cong.,
Court
7
(1950),
294 at 317,
United_States
4.
148 F. 2d 416
(1945)
Whether
5.
understood is
and
J.J.
384 U.S.
,
270
(1966).
,
.
that standard is always wisely applied or
McGowan,
See F.M.
well
Fisher, J.E. Green-
Fplded^_Spindled^_and_Mutilatedi_Ecpr
Mass.:
(Cambridge,
MIT
Press,
.
This
6.
have
may be the reason that the
attempted
(decided 1981)
antitrust
authorities
to invent a doctrine of "shared monopoly."
In_the_Matter_pf_Kellpgg_Company^_et_al J
,
FTC
Docket
No.
See
8883
.
Consideration of the absence of any serious remedy in the
7.
second
Co..,
and by the Brpwn_Shpe
A2umiDum_CpmpaDy_pf_Americaj._et_alj
v.
different matter.
a
1775,
346), as well as in later opinions.
npmic_Analysis_and_y_.S_._yj_.IBM
1983)
(Senate Report No.
United_gtates v. VpD_[s_GrpC£ry_Cp J
3.
Congressional
The same language was
4-5),
pp.
to
(See Senate Report No.
(1914), p. 1.)
was amended in 1950
2nd Sess.
(370 U.S.
wood,
UDited_States, 370 U.S. 294 (1962).
*
Amer ican_Tpbaccp case
328
U.S.
781
(1946))
(Unj.ted_§ tates v.
or of the
Ameiican_Tpbaccp
history of litigation
investigations
in the
blem.
no position on whether structural remedies
I
take
cement industry will illustrate
have been effective in these industries.
27
the
and
prowould
policy towards oligopoly that (unlike the case
tural
tural Section
a
single
all
in
them-
There is no escape from this problem.
it in terms of merger policy either does not solve
else makes it worse in the sense
or
of
it
penalizing
for acts that might later be taken by others.
firms
itself
in
penalized because of actions (also not wrong
Attacking
at
individual firms doing nothing wrong
taken by others.
selves)
struc-
of
cases where all the acts are under the control of
2
firm)
be
will
of course, be considered an objection to a struc-
It may,
8.
two
It is well
to remember that such objections have less force when considering
civil, rather than criminal cases.
United States Department of Justice, Merger_Guidelines
9.
Federal Register, Vol. 49 (1984), 26,284.
United_States v.
10.
U.S. 377
E..
Ij_duPpnt_de_Nemours_and_Cpmpany
See G.W. Stocking and W.F. Mueller,
(1956).
,
353
"The Cello-
Case and the New Competition," American_Ecgnomic_Reyiew 45
phane
(March 1955),
pp.
29-63;
Policy (Cambridge,
Mass.:
and
102;
Posner,
R.A.
C.
Kaysen and D.F. Turner, Antitrust
Harvard University Press,
1959),
p.
Antitr^st_Lawi_AD_Econpmic_Perspectiye
(Chicago and London: University of Chicago Press, 1976), pp. 127128.
11.
the
It may or may not have been a
coincidence that,
after
acquisition was challenged by the Federal Trade Conmmission,
Stouffers began
someone
tasting
"What is it?
a
series of television commercials that
a
featured
Stouffer's product and saying something
It tastes like lasagna, but it isn't lasagna."
case was eventually settled.
28
like
The
12.
For
more detailed discussion of these issues in
a
context of Sherman Act, Section
Chapter
Greenwood, op_._cit_
! ,
13.
F.
295
Amer.ica.t _et_al.1
PP.±_cit..,
of
3.
148 F.
,
2d 416
On these matters,
Chapter
7,
(1945)
at 430.
see Fisher,
McGowan, and Greenwood,
F.M. Fisher and J.J. McGowan,
"On the Misuse
Accounting Rates of Return to Infer Monopoly Profits," Amgiir
Fisher,
Vol.
"On the Misuse of the Profits-Sales Ratio to Infer Mono-
Massachussetts Institute
Power,"
15.
(March 1983), pp. 82-97, and F. M.
73
Paper 364 (Revised April 1986)
Working
.
The HHI is defined as the sum of squares of the
venient decimal points).
of
Technology
of
of the individual firms multiplied by 10,000
HHI
110
,
at 341; Dnited_3tate v. AlumiDum_CpmpaDy_Pf
(1953)
can_Ecpnpmic_Reyiew,
poly
cases, see Fisher, McGowan, and
UDited_States v. United_Shpe_Machine£y_CprppiatipD
Supp.
14.
2
the
shares
(to eliminate incon-
Before the multiplication by 10,000, an
1/n can be thought of as the level of concentration
measured)
that
would occur if there
were
n
(so
identically-sized
firms in the market.
Studies attempting to relate profit levels to HHI values are
not
reliable
competitive
Fisher,
guides to the influence of concentration
behavior.
pp_._cit...
See
Fisher and
McGowan,
on
non-
op J._cit J
and
Even on their own terms, such studies are not
so successful as to warrant basing merger policy on them.
16.
In the Northwest-Republic airline merger
Acguisitipn_Case,
Department
of
Transportation
(NWArRepublic
Docket
43754
(1986)), DOJ insisted that one-stop or connecting airline service
was not in the same market as non-stop service.
29
In so doing,
it
based
arguments on the undeniable fact that
its
travelers
all
non-stop service to one-stop or connecting service if the
prefer
flights leave at the same time and have the same price.
argument
takes
definition.
a
very limited view of substitution
market
and
In fact, people take one-stop or connecting flights
in preference to non-stop flights if they get something
That
Such an
something
lower price.
cally sells for
can
be time-of-day convenience or it can
A large box of
a
thereby.
a
be
a
particular breakfast cereal typi-
higher price than does a small box.
not put them in different markets,
and,
in fact,
That does
the prices of
the different types of flights tend to move together.
For an extended discussion of these matters,
17.
see C. C.
n
vori
lin,
Weizsacker,
Heidelberg,
Barr iers_tp_Entryi_A_Theoretical_Treatment
New York: Springer-Verlag
,
1980)
(Ber-
and Chapter
6
of Fisher, McGowan, and Greenwood, pp_._c.it_..
lpc__cit_.
19.
see Fisher,
For an older one,
18.
McGowan, and Greenwood,
.
[CITE NEEDED].
I
was retained by counsel for Ford who
eventually decided not to bring suit to oppose the joint venture.
20.
Pacj,fic_DiyisipD_Transf ex_Case, Department of Transpor-
tation Docket 43065 (1985).
I
was
a
witness for Northwest
Air-
lines which opposed the acquisition.
My views on the matter are
set forth at length in F. M. Fisher,
"Pan American to United: The
Pacific Division Transfer Case," Massachusetts Institute of Tech-
nology Working Paper 420 (May, 1986)
21.
.
It is worth remarking that the testimony offered by DOJ
in opposition to the acquisition was focussed very heavily on the
30
HHI and the QuideliDSg.
this.
They
purporting
HHIs
had
United and Pan American were ready
fcr
had previously prepared a study for use in rebuttal
to show the not very surprising fact
little
effect on fares.
that
The result was
city-pair
largely
divert argument from the more substantial questions at issue.
31
to
361
UQ75
Date Due
JE26W)
IY 1
OEC
o
'90
^ 1990
dec o
1
ress
Lib-26-67
MIT LIBRARIES
3
TOBO OD5 130 b43